Figure 3
The figure presents a three-dimensional cube diagram illustrating a conceptual framework that categorizes artificial intelligence interaction modes according to digital density, physical complexity, and social presence. The cube is drawn in perspective using solid and dashed lines to indicate visible and hidden edges, with internal dashed grid lines subdividing the structure into eight smaller volumetric sections representing different combinations of the three dimensions. The vertical dimension represents the “Physical realm”, separating levels of complexity. The upper half is labeled “Low complexity”, while the lower half is labeled “High complexity”. The depth dimension represents the “Digital realm”, with density increasing from “Low density” at the front-left region to “High density” toward the back upper area of the cube. The horizontal base dimension represents the “Social realm”, ranging from “High social presence” on the left side to “Low social presence” on the right side. Within the cube, three A I interaction categories are positioned to illustrate their relative operational characteristics. A blue-outlined region labeled “A I-Assisted” occupies the lower-left-front section of the cube, corresponding to high physical complexity, high social presence, and lower digital density, indicating systems that support human decision-making while maintaining strong human involvement. An arrow labeled “A I-Augmented” points diagonally upward from this region toward the center of the cube, representing a transition zone where A I increasingly enhances human capabilities while still retaining meaningful interaction within physical and social contexts. In contrast, a green-outlined region labeled “A I-Automated” appears in the upper-right-back section, corresponding to low physical complexity, high digital density, and low social presence, indicating systems that operate with greater autonomy and reduced direct human interaction.AI configurations in healthcare and the Bolton et al. (2018) cube. Source: Authors' elaboration on Bolton et al. (2018)